Built once, deployed across the portfolio.

Most AI work at PE firms gets stuck at the portco level: each portfolio company runs its own pilots, each pilot dies for the same reasons, and the fund's operating team has no leverage. The right unit of work is the fund, not the portco.

The portco operating model - small teams, short hold periods, EBITDA-led decisions - rewards AI that templates well and resists AI that requires bespoke integration. We optimize for the first.

Finance ops, fraud and controls, procurement, sales intelligence, and cross-portfolio KPI roll-up are the use cases where templating actually works. We have a sharp thesis here and pre-engagement artifacts to back it.

Readiness → Build → Operate, applied at the fund level rather than one portco at a time. Buying moves faster than in regulated industries - the value-creation or operating team can sponsor directly - but the bar is different: whatever we build has to template across the portfolio, not just solve one company's problem. We scope for reuse from the first engagement.

The real gates are data access and hold-period math. Portfolio companies run different systems, so we design for cross-portco data access, segregation, and security up front, and we only commit to use cases whose payback lands comfortably inside the hold period. If a capability cannot be templated, we say so rather than sell a bespoke build.

A typical path is a fund-level Readiness audit to pick the highest-leverage shared use case (finance ops, spend, or KPI roll-up), an 8-14 week Build proven in one portfolio company, then Expand - templating the same capability across the rest of the portfolio where it fits.